Midterm practice
Exam-shaped problems that mix chapters the way a real exam does. Commit to an answer — out loud or on paper — before opening each step. If a problem feels effortful, that's the active learning from Chapter 1 doing its job.
- Benefits and limitations of behavior research and patient studies (L7 Q1, Q2) — and of every brain-measuring technique
- Apply Marr's 3 levels and Tinbergen's 4 questions to movement, flight, vision, and an example of your own
- Understand the effects of damaging each point of the visual system
- Single vs double dissociation, with their weaknesses and fixes
1. Methods
Give three benefits and three limitations, each with an example from the face lectures.
Benefits
- Characterizes representations — orientation-specific vs invariant (Dutch politicians).
- Dissociates representations — faces vs objects (inversion effect).
- Cheap and low-tech.
Limitations
- Sparse, limited data — one response per trial.
- Hard to see fine-grained stages of processing — reaction time lumps perception, decision and motor planning together.
- No relation to the brain.
Use prosopagnosia and patient CK.
Benefits
Evidence that a region is necessary (a causal clue) — an FFA lesion abolishes face recognition. Double dissociations (prosopagnosia vs CK) show two processes are distinct.
Limitations
- Hard to find patients with the "right" lesion.
- Lesions are messy — affect multiple functions.
- Chronic lesions → the brain reorganizes during recovery; effects may reflect recovery, not the lesion.
- Not a direct manipulation — can't turn a region off and back on.
For each question, name the best method and the one limitation that matters most: (a) Is there a region that responds more to tools than to animals? (b) Does the tool response start before or after 200 ms? (c) Is that region necessary for naming tools?
(a) Where?
fMRI — best whole-brain spatial resolution. Key limitation: correlational (can't show causality); also slow and indirect (BOLD).
(b) When?
EEG/MEG (non-invasive, ms timing) — but can't localize to the region. If you need both where and when: ECoG/sEEG, at the cost of rare, invasive, limited-coverage recordings.
(c) Necessary?
Perturbation: patients with lesions there (messy, recovery), or direct electrical stimulation via ECoG (strong causal evidence; can't separate the region from its connections; special patient group).
The takeaway
Where → fMRI; when → EEG/MEG; where + when → ECoG/sEEG; causal → patients/stimulation. And the big-picture sentence: no single method gives the full picture — robust claims come from convergence.
2. Dissociation logic
Percent correct:
| Task A: name tools | Task B: name animals | Task C: name colours | |
|---|---|---|---|
| Controls | 92 | 90 | 99 |
| Group 1 (lesion X) | 45 | 88 | 98 |
| Group 2 (lesion Y) | 89 | 41 | 97 |
What can you conclude, and which task is useless?
Step 1 — What's being asked?
Classify the pattern (none / single / double dissociation) and spot any task that lacks sensitivity.
Step 2 — Tools vs animals
Group 1: tools impaired, animals spared. Group 2: animals impaired, tools spared. Opposite patterns → double dissociation between tool and animal naming, supporting separate (parallel) processes, and ruling out general impairment and task difficulty.
Step 3 — Colours
Everyone is at 97–99%: a ceiling effect. You can't conclude colour naming is unaffected — the task can't detect a deficit.
Serial: Rhythm → Melody → Music. Parallel: Rhythm → Music and Melody → Music. The RH group shows melody impaired / rhythm spared; the LH group shows rhythm impaired / melody spared. Explain which result does the work.
Answer
Melody-impaired / rhythm-spared (RH) fits both models — break the melody stage in either and only melody fails. Rhythm-impaired / melody-spared (LH) is impossible under the serial model — melody depends on rhythm's output. So the LH result is what rules out serial and favours parallel.
3. Frameworks
Give all three Marr levels and all four Tinbergen questions for depth perception.
Marr
Computational: reach and grasp accurately, navigate safely, judge approaching objects. Algorithmic: combine pictorial cues (occlusion, perspective, height, known size, texture, atmosphere, shadows) with non-pictorial cues (disparity, parallax, vergence, accommodation), choosing the interpretation that fits the most cues. Implementation: V1–V5; disparity processed from V1 to V5/MT.
Tinbergen
Function: same as computational. Mechanism: algorithm + implementation above. Development: how depth perception emerges in infancy — e.g. stereopsis needs both eyes aligned during the critical period (strabismus → amblyopia). Evolution: predators (forward-facing eyes, disparity) vs prey (wide field of view); the praying mantis's head-swaying for motion parallax.
4. Clinical reasoning
Patient 1: slow, small movements, resting tremor. Patient 2: involuntary dance-like movements, cognitive decline, family history. Patient 3: a Parkinson's patient on high-dose medication for eight years, now with writhing involuntary movements. For each: diagnosis, pathway, and chain of effects.
Patient 1
Parkinson's. Dopamine loss (substantia nigra) → striatum less active → less inhibition of GPi → more inhibition of thalamus → less excitation of cortex. Go pathway.
Patient 2
Huntington's (genetic). Striatal loss in the no-go pathway → less inhibition of GPe → more inhibition of STN → less excitation of GPi → less inhibition of thalamus → chorea. Plus dementia.
Patient 3
L-dopa-induced dyskinesia: unnatural dopamine influx + receptor sensitization → small amounts of dopamine now trigger involuntary movement. Option: DBS of GPi, which releases the thalamus directly without extra dopamine.
(a) Loses the left half of the visual field in both eyes. (b) Loses the outer half of each eye's field. (c) Blind in the right eye only. (d) Fluent nonsense speech, poor comprehension. (e) Can't recognize her husband's face but recognizes his voice and names objects fine. (f) Sees moving cars as a series of snapshots.
Answers
(a) Right optic tract / LGN / radiation / V1 — after the chiasm (contralateral homonymous hemianopia). (b) Optic chiasm (bitemporal hemianopia). (c) Right optic nerve, before the chiasm. (d) Wernicke's area (Wernicke's aphasia). (e) FFA — prosopagnosia. (f) MT/V5 — akinetopsia (patient LM).
(a) A retired linebacker with depression, explosive temper and memory complaints. (b) A cyclist in an immediate coma after a crash with a violent head twist. (c) Sudden worst-ever headache, vomiting, collapse. (d) A clot in the anterior cerebral artery.
Answers
(a) CTE stage 2 — repeated head trauma, frontal tau. (b) Diffuse axonal injury — rotational shear. (c) Hemorrhagic stroke from a ruptured aneurysm (treat unruptured ones by clipping or coiling). (d) Ischemic stroke in frontal + parietal territory.
5. Vision mechanisms
Explain the grey dots using ON-center, OFF-surround ganglion cells at an intersection vs a corridor.
Model answer
Both cells' centers sit on white and are equally excited. At an intersection, white extends in four directions, so more light hits the OFF-surround than in a corridor, where white runs in only two. More surround inhibition → weaker response → the intersection is seen as darker, a grey dot.
Connect: not seeing your retinal blood vessels; tilt aftereffect; waterfall (motion) aftereffect.
Model answer
All three come from neurons reducing their response to sustained stimulation. Blood vessels: unchanging input is filtered out (visual adaptation). Tilt: fatigued orientation-tuned simple cells bias the population away from the adapting tilt. Motion: fatigued direction-tuned MT cells let the opposite direction dominate, so a static scene seems to move the opposite way. The brain codes change and balance, not absolute input.
6. Foundations
Explain how each raises effective neurotransmitter signalling: SSRIs, dopamine agonists, MAO-B inhibitors.
Answer
SSRIs: block serotonin reuptake → more serotonin stays in the synapse. Dopamine agonists: mimic dopamine and bind its receptors directly. MAO-B inhibitors: block an enzyme that breaks dopamine down → it lasts longer.